In [1]:
import cv2
import numpy as np
import matplotlib.patches as patches
from IPython.display import clear_output
import matplotlib.pylab as plt
import os
try:
import google.colab
IN_COLAB = True
%matplotlib inline
from google.colab import drive
from google.colab.patches import cv2_imshow
drive.mount('/content/gdrive')
DRIVE_ROOT='/content/gdrive/My Drive/opencv/'
!pip install face_recognition
except:
IN_COLAB =False
%matplotlib notebook
DRIVE_ROOT='./'
import face_recognition
face_dir = os.path.join(DRIVE_ROOT,'known_face')
OUT_DIR =os.path.join(DRIVE_ROOT,'output')
In [3]:
# detect face locations
def detect_face_locations(image,do_gray=False,do_scale=False):
if do_scale:
small_frame = cv2.resize(image, (0,0), fx=0.25, fy=0.25)
else:
small_frame = image
if do_gray:
gray = cv2.cvtColor(small_frame,cv2.COLOR_BGR2RGB)
rgb_frame = gray[:,:,::-1]
else:
rgb_frame=small_frame[:,:,::-1]
# face_locations = face_recognition.face_locations(rgb_frame,model="cnn")
face_locations = face_recognition.face_locations(rgb_frame)
for face_location in face_locations:
(top,right,bottom,left)=face_location;
if do_scale:
top *= 4
right *= 4
bottom *= 4
left *= 4
cv2.rectangle(image, (left,top), (right,bottom), (0,0,255),2)
#plt.plot(face_location[1],face_location[0],'bo')
#plt.plot(face_location[1],face_location[2],'bo')
#plt.plot(face_location[3],face_location[2],'bo')
#plt.plot(face_location[3],face_location[0],'bo')
#plt.show()
return image
def detect_face_save(image):
rgb_frame=image[:,:,::-1]
height,width = image.shape[:2]
#print('width',width,'height',height)
# face_locations = face_recognition.face_locations(rgb_frame,model="cnn")
face_locations = face_recognition.face_locations(rgb_frame)
n=len(face_locations)
cols=5
if n<=cols:
cols=n
rows=1
else:
cols=5
rows=int(1+n/3)
r=1
c=1
i=1
#print(rows,cols)
fig = plt.figure(figsize=(cols*3,rows*3))
for face_location in face_locations:
print(face_location)
(top,right,bottom,left)=face_location;
#print('left',left,'top',top,'right',right,'bottom',bottom)
w = int((right - left)/2)
h = int((bottom - top)/2)
print('w',w,'h',h)
top=max(0,top-h-h)
bottom = min(height-1,bottom+h)
left=max(0,left-w)
right=min(width-1,right+w)
#print('left',left,'top',top,'right',right,'bottom',bottom)
image_face=image[top : bottom, left : right]
#cv2.rectangle(image, (left,top), (right,bottom), (255,0,0),2)
img = cv2.cvtColor(image_face,cv2.COLOR_BGR2RGB)
ax = fig.add_subplot(rows,cols,i)
ax.imshow(img)
c+=1
if c> cols:
c=0
r+=1
# plt.imshow(img)
# plt.show()
i+=1
plt.show()
img = cv2.cvtColor(image,cv2.COLOR_BGR2RGB)
plt.imshow(img)
plt.show()
image=cv2.imread('test-data/IMG_5710.JPG')
detect_face_save(image)
#image=detect_face_locations(image)
#cv2.imshow('show',image)
#if cv2.waitKey(0) & 0xFF:
# cv2.destroyAllWindows()
img = cv2.cvtColor(image,cv2.COLOR_BGR2RGB)
plt.imshow(img)
plt.show()
In [12]:
known_face_encodings=[]
known_face_names=[]
subject_image_names=os.listdir(face_dir)
for image_name in subject_image_names:
if image_name.startswith("."):
continue
name=os.path.splitext(image_name)[0];
image_path = os.path.join(face_dir,image_name)
image = face_recognition.load_image_file(image_path)
face_encoding = face_recognition.face_encodings(image)[0]
known_face_encodings.append(face_encoding)
known_face_names.append(name)
print(name)
detect_face_locations(image)
#img = cv2.cvtColor(image,cv2.COLOR_BGR2RGB)
plt.imshow(image)
plt.show()
#if IN_COLAB:
# cv2_imshow(image)
#else:
# cv2.imshow('training',image)
# if cv2.waitKey(500) & 0xFF:
# cv2.destroyAllWindows()
In [4]:
# single frame recognition
def recognize_face(frame,do_scale=False,threshold=0.5,debug=False):
face_locations=[]
face_encodings=[]
face_names=[]
if do_scale:
# resize frame to 1/4 size for faster processing
small_frame = cv2.resize(frame, (0,0), fx=0.25, fy=0.25)
else:
small_frame = frame
result=False
# convert the image from BGR color to RGB color
rgb_small_frame = small_frame[:,:,::-1]
# find all the faces and face encodings in the frame
face_locations = face_recognition.face_locations(rgb_small_frame)
#if debug:
#print(face_locations)
face_encodings = face_recognition.face_encodings(rgb_small_frame,face_locations)
result=False
for face_encoding in face_encodings:
# see if the face is a match for the known face(s)
matches = face_recognition.compare_faces(known_face_encodings, face_encoding)
name = 'Unknown'
# use the known face with the smallest distance
face_distance = face_recognition.face_distance(known_face_encodings,face_encoding)
best_match_index = np.argmin(face_distance)
if matches[best_match_index]:
if face_distance[best_match_index]<threshold:
name = known_face_names[best_match_index]
else:
name = '?'+known_face_names[best_match_index]+'?'
if debug:
print(name,face_distance)
if name == 'Toma':
result=True
face_names.append(name)
#display result
for(top,right,bottom,left),name in zip(face_locations,face_names):
#scale back up face locations since the frame detected in was scaled to 1/4 size
if do_scale:
top *= 4
right *= 4
bottom *= 4
left *= 4
#draw a box around the face
cv2.rectangle(frame, (left,top), (right,bottom), (0,0,255),2)
# draw a label with a name below the face
font = cv2.FONT_HERSHEY_PLAIN
font_scale = 1.2
thickness = 2
margin=1
size = cv2.getTextSize(name, font, font_scale, thickness)
text_width = size[0][0]
text_height = size[0][1]
cv2.rectangle(frame, (left, bottom), (left+text_width+margin, bottom+text_height+margin), (0, 0, 255), cv2.FILLED)
cv2.putText(frame, name, (left,bottom + text_height), font,font_scale, (255,255,255), thickness)
return result,frame
In [5]:
# test functions
# format time(second) as HH:MM:ss.mmm
def format_time(sec):
msec = (int)(sec * 1000.)
s, ms = divmod(msec, 1000)
m, s =divmod(s,60)
return '{:02d}:{:02d}.{:03d}'.format(m,s,ms)
def format_time_forfile(sec):
msec = (int)(sec * 1000.)
s, ms = divmod(msec, 1000)
m, s =divmod(s,60)
return '{:02d}m{:02d}s{:03d}ms'.format(m,s,ms)
# single image
def single_image(filename,detect_only=False,debug=False):
#image=cv2.imread("img_interviewMovie.jpg")
image=cv2.imread(filename)
if detect_only:
image=detect_face_locations(image)
else:
ret,image=recognize_face(image,debug=debug)
img = cv2.cvtColor(image,cv2.COLOR_BGR2RGB)
plt.imshow(img)
plt.show()
file_name=os.path.join(OUT_DIR,os.path.basename(filename))
cv2.imwrite(file_name,image)
if IN_COLAB:
cv2_imshow(image)
else:
cv2.imshow("sigle test",image)
if cv2.waitKey(0) & 0xFF:
cv2.destroyAllWindows()
def getInterval(fps):
interval =1.0/fps
time_wait = (int)(interval * 1000.0)
print('interval: ',interval,'time_wait: ',time_wait)
return interval,time_wait
def video_proc(filename,output=None,capture=False,detect_only=False,no_wait=False,view=True):
if capture:
pre = os.path.splitext(os.path.basename(filename))[0]
cap = cv2.VideoCapture(filename)
width = int(cap.get(cv2.CAP_PROP_FRAME_WIDTH))
height = int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
fps = cap.get(cv2.CAP_PROP_FPS)
interval,time_wait = getInterval(fps)
time_stamp=0.0
frame_count=0
if no_wait:
time_wait=1
font = cv2.FONT_HERSHEY_SIMPLEX
font_scale = 1.2
thickness = 2
margin=1
size = cv2.getTextSize(name, font, font_scale, thickness)
text_width = size[0][0]
text_height = size[0][1]
counter = 0
# VideoWriter を作成する。
if output is not None:
if IN_COLAB:
fourcc = cv2.VideoWriter_fourcc(*"MJPG")
else:
fourcc = cv2.VideoWriter_fourcc(*'DIVX')
writer = cv2.VideoWriter(os.path.join(OUT_DIR,output), fourcc, fps, (width, height))
else:
writer = None
if capture:
CAPTURE_DIR=os.path.join(OUT_DIR,pre)
os.makedirs(CAPTURE_DIR,exist_ok=True)
while cap.isOpened():
ret,frame = cap.read()
if not ret:
cap.release()
break
frame_count += 1
if detect_only:
frame=detect_face_locations(frame)
else:
result,frame=recognize_face(frame)
cv2.putText(frame,format_time(time_stamp),(margin,margin+text_height),font,font_scale,(255,0,0),thickness)
if result:
counter += 1
print(str(counter),format_time(time_stamp))
if capture:
out_name=os.path.join(CAPTURE_DIR,pre+'_'+format_time_forfile(time_stamp)+'.png')
cv2.imwrite(out_name,frame)
if not IN_COLAB & view :
cv2.imshow('video',frame)
k = cv2.waitKey(time_wait) & 0xff
if k == 27:
break;
if writer is not None:
writer.write(frame)
time_stamp += interval
if not IN_COLAB & view:
cv2.destroyAllWindows()
if writer is not None:
writer.release()
cap.release()
In [ ]:
image_name="./test-data/IMG_5710_S.JPG"
#image_name="./test-data/IMG_6109.JPG"
#image_name=os.path.join(DRIVE_ROOT,"img_interviewMovie.jpg")
#image_name=os.path.join(DRIVE_ROOT,"test-data/12.jpg")
#single_image(image_name,detect_only=False,debug=True)
image=cv2.imread(image_name)
detect_face_save(image)
In [10]:
import time
# name='gassou_15sec_small'
# name='gassou_full_small'
name='gassou_full_large'
video_name=os.path.join(DRIVE_ROOT,name+'.mp4')
out_name = name+'.avi'
start = time.time()
video_proc(video_name, output=out_name, capture=True, detect_only=False, view=True)
#video_proc(0, detect_only=False, view=True)
elapsed_time=time.time() - start
print('elapsed time: ',format_time(elapsed_time))
In [0]:
if IN_COLAB:
file_name = os.path.join(OUT_DIR,out_name)
#file_name ="/content/gdrive/My Drive/opencv/output/gassou_full_large/gassou_full_large_04m*.png"
from google.colab import files
files.download(file_name)
In [0]:
import glob
#image_names=glob.glob(os.path.join(OUT_DIR,'gassou_full_large/*04m27s*.*'))
#image_names=glob.glob(os.path.join(OUT_DIR,'gassou_full_large/*.*'))
image_names=[]
image_names.append(os.path.join(OUT_DIR,'gassou_full_large/gassou_full_large_03m14s068ms.png'))
image_names.append(os.path.join(OUT_DIR,'gassou_full_large/gassou_full_large_04m46s828ms.png'))
image_names.append(os.path.join(OUT_DIR,'gassou_full_large/gassou_full_large_06m23s966ms.png'))
print('count:',len(image_names))
#image_names_sub=image_names[51:60]
for image_name in image_names:
print(os.path.basename(image_name))
image=cv2.imread(image_name)
cv2_imshow(image)
In [ ]:
In [ ]:
In [ ]: